Douglas Rudeen


At the time of this writing, the Supreme Court seems to have abandoned establishing an objective test for drawing nonpartisan districts altogether—leaving this task largely to state-level courts and legislatures in the aftermath of Rucho v. Common Cause.1 Many have seen this deferment to the states as the latest in a series of unsatisfactory ‘balks,’ and have openly wondered what redistricting laws will look like in the next several years as a result of the Court’s general refusal to intervene in this area outside of Voting Rights Act2 litigation. This paper will argue that there is at least one foreseeable outcome— that (in the absence of mathematically rigorous legal standards for gerrymandering) state-level redistricting processes will be left vulnerable to abuse through the use of sophisticated artificial intelligence platforms. To make this argument, the paper will first provide a brief and beginner-friendly primer on the basics of AI and machine learning. Then it will detail how AI is uniquely suited to perpetrate gerrymanders in ways that computer systems would not have been able to during the 2010 redistricting cycle. Finally, it will conclude with a policy recommendation—that the only reliable way to forestall gerrymandering in the age of AI is to employ a form of fully-automatic redistricting, or a novel semi-automatic redistricting process I will call the “Rawlsian Default.”